Information Technology Reference
In-Depth Information
In this paper, we present a case study by combining EEG/ERP and ACT-R
for investigating human computation mechanism. In particular, we focus on two
digits addition tasks with two diculty levels: with or without carry, and sys-
tematically perform a set of behavior and EEG experiments, as well as with the
help of ACT-R modeling, to explore the cognitive processes and their neural ba-
sis. The rest of the paper is organized as follows. Section 2 provides background
and related work on cognitive neuroscience, human thinking process and ACT-R
research. Sections 3 explains how to design the experiments of an ERP mental
arithmetic task. Sections 4 describes how to do ERP/topography data analyses
and ACT-R simulation, respectively, as an example to investigate human com-
putation mechanism and to show the usefulness of combining ERP and ACT-R.
Finally, Sections 5 gives concluding remarks.
2
Background and Related Work
Generally speaking, capabilities of human intelligence can be broadly divided
into two main aspects: perception and thinking. So far, the main disciplines with
respect to human intelligence are cognitive psychology that mainly focuses on
studying mind and behavior with respect to “thinking oriented” high cognitive
functions and models, as well as neuroscience that mainly focuses on studying
brain and biological models of intelligence. In cognitive neuroscience, although
many advanced results with respect to “perception oriented” study have been
obtained, only a few of preliminary, separated studies with respect to “thinking
oriented” and/or a more whole information process have been reported [3,6,14].
Furthermore, an integrated interpretation with those results from both cognitive
psychology and cognitive neuroscience is needed.
ACT-R, one of the best known, fully implemented, and free to public cogni-
tive architecture models, is a theory on how the structure of brain achieves the
function of adaptive cognition in a system level, as well as a platform to build
computational models to simulate/predict human cognitive behavior, including
a wide cognitive problems. It can perform complex dynamic tasks, which usually
emphasize perceptual-motor components and their coordination with other cog-
nitive components (learning and memory, reasoning, and so on) and with strong
time pressure [1,2,13].
In recent years, a series of fMRI experiments have been performed to explore
the neural basis of cognitive architecture and to build a two-way bridge between
the information processing model and fMRI [11,12]. The patterns of the activa-
tions of brain areas corresponding to the buffers of the major modules in ACT-R
are highly consistent across these experiments; and ACT-R has successfully pre-
dicted the Blood Oxygenation Level-Depend (BOLD) effect in these regions. A
recent result is to use the imaging data to identify mental states as the student is
engaged in algebraic problems solving, in which fMRI is used to track the learn-
ing of mathematics with a computer-based algebra tutor and cognitive models
in the ACT-R architecture is used to interpret imaging data [1].
